Sarah John, Ph.D. FairVote: The Center for Voting and Democracy 6930 Carroll Avenue, Suite 610, Takoma Park, Maryland

RANKED CHOICE VOTING CIVILITY PROJECT RESEARCH REPORT 4, APRIL 2015 Results of the Rutgers-Eagleton Institute of Politics poll on voter perceptions and experiences with ranked choice voting in November 2014 city elections in California Sarah John, Ph.D. FairVote: The Center for Voting and Democracy 6930 Carroll Avenue, Suite 610, Takoma Park, Maryland 20912 www.fairvote.org Caroline Tolbert, Ph.D. Department of Political Science University of Iowa Caroline-tolbert@uiowa.edu

Contents 1. RCV in the Bay Area: An Introduction... 3 2. Major Findings... 5 3. Voter Perceptions of Local Elections, Candidates and Campaign Tone... 7 3.1 Satisfaction with the Conduct of Campaigns... 8 3.2 Candidate Criticism... 11 3.3 Perceived Negativity... 14 4. Voter Understanding of Ballot Instructions, Voting Systems & the Top-Two Primary. 18 4.1 Ease of Understanding Voting Instructions... 18 4.2 Understanding RCV and Plurality... 21 4.3 Understanding RCV and Top-Two... 25 5. Support for RCV in Californian Cities... 27 6. Conclusion... 30 Appendix: Methodology... 31 Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 2

1. RCV in the Bay Area: An Introduction Ranked choice voting (RCV) a voting system in which voters rank candidates in order of preference, usually their top three is used to conduct municipal elections in four California Bay Area cities, as well as local elections in other U.S. cities including Minneapolis and St. Paul, Minnesota and Cambridge, MA. In California, San Francisco was the first city to switch to RCV, adopting the system to elect all city officials by a charter amendment in 2002 and holding its first RCV elections in 2004. In 2006, Oakland voters passed a charter amendment to adopt RCV for city officials, with 69% of voters in favor of making the switch. In 2010, Berkeley and San Leandro followed suit, embracing ranked choice voting to elect their city officials. In November of 2010, Oakland, Berkeley, and San Leandro all held their first RCV elections, with Oakland s highly competitive mayoral election receiving national media attention. In 2011, many of San Francisco s RCV races, including Mayor and several Board of Supervisors districts, were extremely competitive. And in 2012 Oakland, Berkeley, and San Leandro used RCV for a second time, with competitive races for Berkeley s mayor, and city council positions in Oakland and San Leandro. In November 2014, all four cities used RCV to elect city mayors, city council members, school district seats and city-wide executive positions: 24 offices in total, though not all were competitive (Table 1). As the experience of voters, candidates and electoral administrators with RCV mounts, researchers can now study whether changing from winner-take-all (also known as plurality voting ) to ranked choice voting elections encourages the development of a more positive campaign dialogue and a more civil politics. In theory, RCV offers incentives for candidates to cooperate and campaign civically, since each vote is not an all or nothing battle and candidates can appeal to strong supporters of other candidates for their second or third choices. In fact, candidates often need a combination of first choice rankings, as well as some lower rankings, in order to win RCV elections. To explore the experiences of voters with RCV elections, the Rutgers-Eagleton Institute of Politics has conducted two polls one in 2013 and another in 2014 that explore the impact of RCV on city elections in the United States. As part of a broader project, funded by the Democracy Fund, these surveys were designed and analyzed by Professors Caroline Tolbert (University of Iowa) and Todd Donovan (Western Washington University). Kellen Gracey (University of Iowa) also contributed to the data analysis for this study. Each survey asked a random sample of more than 2,400 likely voters about their perceptions of local campaigns and their understanding of ballot instructions and voting systems. The great majority of respondents had voted in their local election that year. The surveys were conducted in English and Spanish and on cell and landline telephones. The survey methodology is expounded in Appendix 1. Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 3

Table 1: Bay Area Offices Filled Using RCV in November 2014 1 Number of Candidates Percentage of first choice votes of winner (competitive races italicized) Berkeley City Auditor 1 98.34% Member, City Council, Dist. 1 3 40.63% Member, City Council, Dist. 4 1 95.40% Member, City Council, Dist. 7 2 55.50% Member, City Council, Dist. 8 4 29.15% Oakland Mayor 16 29.48% City Auditor 2 76.34% Member, City Council, Dist. 2 5 35.93% Member, City Council, Dist. 4 3 63.95% Member, City Council, Dist. 6 5 42.60% School Director, District 2 2 75.44% School Director, District 4 4 38.21% School Director, District 6 2 58.02% San Francisco Assessor 1 98.05% Public Defender 1 98.26% Supervisor, Dist. 2 2 78.61% Supervisor, Dist. 4 1 96.70% Supervisor, Dist. 6 4 67.42% Supervisor, Dist. 8 5 77.67% Supervisor, Dist. 10 5 46.49% San Leandro Mayor 4 48.46% Member, City Council, Dist. 1 4 42.23% Member, City Council, Dist. 3 3 56.67% Member, City Council, Dist. 5 3 53.54% In this report, Drs. Sarah John (FairVote) and Caroline Tolbert (University of Iowa) present the results from the November 2014 Rutgers-Eagleton Institute of Politics poll of likely voters in 11 Californian cities: four cities holding elections with RCV (Berkeley, Oakland, San Francisco and San Leandro), and seven control cities, 1 which held elections under plurality (winner-take-all) rules. Demographic factors were comparable in the RCV and non-rcv cities. The report focuses on voter experience with, and support of, RCV, broken down by demographic factors and socioeconomic status. The next report in the Ranked Choice Voting Civility Project Research Report series will explore the geography of voter experience with, and support of, RCV. 1 Alameda, Anaheim, Richmond, San Jose, Santa Ana, Santa Clara and Stockton Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 4

2. Major Findings Voter Perceptions of Local Elections, Candidates and Campaign Tone Likely voters in cities that used ranked choice voting (RCV) in their local elections were more satisfied with the conduct of candidate campaigns, and perceived less candidate criticism and negative campaigning in the lead up to the November 2014 elections. In the RCV cities of Berkeley, Oakland, San Francisco and San Leandro, only 53 percent of respondents remembered candidates criticizing each other, compared to 65 percent in plurality cities. Similarly, more respondents in cities using RCV (17%) reported reduced negativity in local election campaigns than in cities that without RCV (12%). Virtually every demographic group studied including less affluent respondents, college graduates, Latinos, African Americans, women, Independents and unmarried people reported less negativity and less candidate criticism in RCV cities than in plurality cities. Independent voters in RCV cities were more satisfied with candidates campaigns: Respondents in RCV cities expressed significantly higher levels of satisfaction with candidates campaign conduct in the 2014 local elections than did their counterparts in cities using winner-take all elections. In plurality cities, less than 43% of Independents were satisfied, as opposed to 53% of Democrats and 55% of Republicans. In RCV cities, there was no statistically significant difference between the reported satisfaction of Democrats (52%), Republicans (50%) and Independents (50%). The dissatisfaction of Independents with campaigns in plurality elections may suggest that plurality elections encourage more ideologically extreme campaigns, even in non-partisan local elections. Voter Understanding of Ballots Instructions, Voting Systems and the Top- Two Primary In all cities surveyed, an overwhelming majority of voters found the ballot instructions easy to understand. Eighty-eight percent of voters in RCV cities found their RCV ballot instructions easy to understand. This was only slightly lower than in plurality cities, where 93% of voters found their plurality ballot instructions easy to understand. These findings were consistent across demographic groups. Self-reported understanding of RCV is high and compares favorably to understanding of plurality and the Top-Two primary. The percentage of voters in RCV cities who understood RCV somewhat or very well (84%) was equivalent to the percent of voters in plurality cities who understood plurality voting (83%). More respondents (49%) in RCV cities reported understanding RCV extremely or very well than reported understanding the top-two primary extremely or very well (40%). Continued Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 5

African-American voters were much more likely to find RCV ballot instructions easy to understand: Ninety percent of African-American voters in RCV cities found the ballot instructions easy to understand, compared to an abysmal 65 percent in plurality cities. An equivalent percent of African-American respondents reported understanding RCV in RCV cities (88%) as understood plurality in plurality cities (86%). This suggests that understanding of ballot instructions is more about the careful design of instructions than it is about which voting system a city employs. Voter Support for Ranked Choice Voting Ranked choice voting receives high overall support: Among all likely voters with an opinion about RCV, 57 percent in the four Bay Area cities that use RCV agreed that ranked choice voting, where voters can rank candidates in order of preference with their first choice counting most, should be used in local elections in their city. A majority backed RCV in each city, including 60 percent in Oakland. Even in cities that do not use RCV, a majority of likely voters (54%) supported RCV. RCV support greatest among people of color, young people, and low-income voters: While a majority of most demographic groups supported RCV, the strongest support for RCV came from the young under age 30 (61%), those with a family income under $40,000 (63%), individuals who did not attend college (65%) as well as Asian American (72%) and Latino (59%). The findings of the 2014 study into voter perceptions of, and experiences with, RCV in local elections in California are consistent with results from the 2013 study written by Todd Donovan, Caroline Tolbert and Kellen Gracey, conducted by the Rutgers-Eagleton Institute of Politics in Midwest and East coast cities. More information is available at: www.fairvote.org/research-and-analysis/local-elections/ranked-choice-voting-civilityproject. This report is the fourth in the Ranked Choice Voting Civility Project Research Report series. Other reports in this series, available at the project's website, include: John, Sarah. February 2015. Ranked Choice Voting in Practice: Content Analysis of Campaign Tone in Newspapers and Twitter Feeds in 2013 RCV Elections. Ranked Choice Voting Civility Project Research Report #3 John, Sarah. February 2015. Ranked Choice Voting in Practice: Candidate Civility in Bay Area Elections, November 2014. Ranked Choice Voting Civility Project Research Report #2. Douglas, Andrew. April 2014. Ranked Choice Voting and Civility: New Evidence from American Cities. Ranked Choice Voting Civility Project Research Report #1. For more information, visit: www.fairvote.org/research-and-analysis/local-elections/rankedchoice-voting-civility-project/ Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 6

3. Voter Perceptions of Local Elections, Candidates and Campaign Tone: With and Without RCV In the November 2014 Rutgers-Eagleton Institute of Politics poll, directed by Dr. David Redlawsk, asked likely voters three key questions about the tone and conduct of campaigns for local offices in their city in 2014. Respondents were asked whether they were 1) satisfied with the conduct of candidate campaigns; 2) how often they remembered candidates criticizing each other; and 3) whether the campaigns were more or less negative in 2014 compared to previous local elections. We first present the findings combining respondents from RCV and non-rcv cities to understand overall patterns and develop a baseline. We then report results for respondents from RCV and non-rcv cities separately for the comparison analysis. In RCV cities, respondents reported higher levels of satisfaction, lower levels of candidate criticism and less negative campaigning than in plurality cities. 2 These tendencies were especially strong with regards to candidate criticism and negative campaigning. All socioeconomic and demographic groups in RCV cities with the sole exception of those who had attended some college (but not completed their degree) reported lower perceptions of candidate criticism than did the same demographic groups in plurality cities (though not all the differences were statistically significant). In RCV cities, a higher proportion of respondents in most socioeconomic and demographic groups including African Americans, Latinos and Asian Americans reported noticing lower levels of negativity in 2014 compared to prior local contests. Indeed, only two groups reported a less campaign negativity in plurality cities than in RCV cities: those who did not attend college and self-reported Republicans. These findings are consistent with the 2013 Rutgers-Eagleton poll, which found that respondents in the RCV cities of Minneapolis, MN, St. Paul, MN, and Cambridge, MA, were more satisfied with candidate conduct, and perceived less candidate criticism and negativity than did respondents in the seven plurality control cities (including Seattle, WA, Tulsa, OK and Boston, MA). 3 In the 2013 poll, 42% of RCV city respondents reported that local campaigns were less negative than previous local contests. Only 28% of respondents in plurality control cities reported less negativity in 2013. 2 The seven plurality control cities in 2014 were Alameda, Anaheim, Richmond, San Jose, Santa Ana, Santa Clara and Stockton. 3 For more information see: Andrew Douglas (April 2014) Ranked Choice Voting and Civility: New Evidence from American Cities Ranked Choice Voting Civility Project Research Report No.1. Available at: http://www.fairvote.org/assets/rcv-civility-project/ranked-choice-voting-civility-study-april-2014.pdf. The seven plurality control cities in 2013 were Boston, Lowell and Worcester in Massachusetts, Des Moines and Cedar Rapids in Iowa, Seattle in Washington and Tulsa in Oklahoma. Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 7

In non-rcv cities, Independents (compared to Democrats or Republicans) were especially unsatisfied with candidate conduct under plurality elections and they reported higher levels of candidate criticism and campaign negativity. These findings may be related to the use of plurality voting and the incentives it provides to candidates to mobilize their most fervent supporters with harsh attacks on competing viewpoints and opposing candidates. Interestingly, Asian Americans reported the lowest levels of satisfaction with the conduct of campaigns of any group in RCV cities. Because of the low number of Asian respondents in the survey, this result is not statistically significant. Additionally, it is important to note that Asian-American political participation in California is relatively low with turnout levels in 2012 (65% of registered voters) around seven percentage points lower than the average (72% of registered voters). 4 Given this, we would expect lower levels of political knowledge and satisfaction with government. However, as will be shown below, Asian Americans reported high levels of support for RCV. 3.1 Satisfaction with the Conduct of Campaigns Overall Patterns Respondents were asked: In general, are you very satisfied, fairly satisfied, not very satisfied, or not at all satisfied with the way most candidates have conducted their campaigns in the recent local election? Their responses reveal that likely voters, especially Independents, were more satisfied in RCV cities than plurality cities. Overall, just under half (47%) of respondents in all cities reported being very or fairly satisfied with the way most candidates conducted their campaigns in the recent local election. This is the baseline for comparison, as Figures 1 and 2 include all respondents from RCV and non- RCV cities. There was little variation among age groups, with young voters (under 30 years of age) only slightly more likely to be satisfied than their older counterparts (Figure 1). However, significantly fewer respondents (35%) without any college education were satisfied with the conduct of candidates than were respondents who had attended college (50%). Since education is associated with increased electoral participation, this pattern is consistent with expectations. Across demographic groups, Asian Americans respondents reported lower levels of satisfaction with candidates campaign conduct (Figure 2), with 42% reporting that they were very or somewhat satisfied with the conduct of campaigns. As discussed above, political participation rates of Asian Americans tend to be lower than for African Americans or White non-hispanic adults, which may spill over into their evaluation of campaigns and elections. In any case, Asian respondents did not report statistically significantly different levels of satisfaction than Black or Latino respondents. 4 Asian Americans Advancing Justice - Los Angeles (2013) Asian Americans at the Ballot Box: The 2012 General Election in California. Available at: http://advancingjusticela.org/sites/default/files/aa_ballot_box_ca_2012_final.pdf Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 8

Male respondents were more likely than female respondents to report satisfaction (50% to 45% respectively). Perhaps unsurprisingly, given the results of the 2014 election nationally and in Congress, respondents who identified with the Republican Party reported higher levels of satisfaction (54%) than did Democrats (46%) or Independents (45%). Figure 1: Satisfaction with the Conduct of Campaigns, by Age, Education and Income, All Surveyed Cities Figure 2: Satisfaction with the Conduct of Campaigns, by Select Demographics, All Surveyed Cities Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 9

Comparing RCV and non-rcv Cities In both RCV and plurality cities (designated Non-RCV cities in the figures), a greater proportion of more educated and higher income respondents reported being satisfied with the conduct of campaigns than did less educated and lower income respondents (Figure 3). In RCV cities, reported levels of satisfaction were modestly higher (50%) than in plurality cities (46%). Across age groups, satisfaction with the conduct of elections was slightly higher in RCV than non-rcv cities. The relationship between income and satisfaction was slightly stronger in RCV cities than plurality cities. Half of likely voters in RCV cities were very or fairly satisfied with the way most candidates conducted their campaigns in 2014 compared to fortysix percent in plurality cities. Figure 3: Satisfaction with the Conduct of Campaigns, by Age, Education and Income, All Surveyed Cities In plurality cities, just 43% of Independents were satisfied with the conduct of candidates, compared to 53% of Democrats and 55% of Republicans. A key unsatisfied group in plurality cities was likely voters who were unaffiliated with any political party: in plurality cities, 43% of Independent respondents reported being satisfied with the conduct of the 2014 local campaign, as opposed to 53% and 55% of Democrats and Republicans, respectively (Figure 4) a difference significant at the 90% confidence level. In RCV cities, there was no statistically significant difference between the reported satisfaction of Democrats (52%), Republicans (50%) and Independents (50%). The dissatisfaction of Independents with campaigns Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 10

in plurality elections may indicate that winner take all elections tend to encourage more ideologically extreme campaigns, even in non-partisan local elections. In RCV cities, African American and Asian Americans were less likely to be satisfied with the conduct of candidates campaign than where White and Latino respondents (Figure 4). While Asian Americans reported lower levels of satisfaction than other race and ethnic groups in RCV cities and their counterparts in plurality cities, their small sample size means that these differences are not statistically significant. Figure 4: Satisfaction with the Conduct of Campaigns, by Select Demographics, RCV and Plurality Cities 3.2 Candidate Criticism Overall Patterns The data indicate that RCV is associated with more civil discourse between candidates in local elections. Respondents were asked in the recent election, how much time would you say the candidates spent criticizing their opponent? Across all surveyed cities, 60% of respondents reported remembering candidates criticizing other candidates either a great deal or some of the time. In RCV cities, respondents remembered candidates criticizing other candidates less often than respondents in plurality cities. Indeed, there was a full 12 percentage point gap, with only 53% of respondents in RCV cities remembering criticism compared to 65% of plurality city respondents. In plurality cities, 65% of likely voters remember candidates criticizing each other a great deal or some of the time compared to 53% in RCV cities. Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 11

Figure 5: Remember Candidates Criticizing Each Other, by Age, Education and Income, All Surveyed Cities Figure 6: Remember Candidates Criticizing Each Other, by Select Demographics, All Surveyed Cities Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 12

Comparing RCV and non-rcv Cities Figures 5 and 6 indicate that across all cities RCV and plurality surveyed, higher income, Latino and male respondents were more likely to recall candidates criticizing each other than other demographic groups. Around 65% of Latinos and individuals with family income in excess of $75,000 per year remembered local election candidates criticizing each other a great deal or some of the time. All demographic groups studied reported lower or statistically equivalent levels of candidate criticism in RCV cities compared to plurality cities. When we compare responses in RCV and plurality cities, we see that the every demographic group except one reported lower or statistically equivalent levels of candidate criticism. Older respondents (aged 65 years and over) were especially likely to have remembered less criticism in RCV cities (43%) than in plurality cities (67%) (Figure 7). Figure 7: Remember Candidates Criticizing Each Other, by Age, Education and Income, RCV and Plurality Cities As well as improving the experience of older voters, RCV appears to improve the tone of elections in the recollections of unmarried people, women, Latinos and African Americans, and less-educated citizens. These groups reported particularly low levels of candidate criticism relative to their counterparts in plurality cities (Figure 8). All partisan groups Democrats, Independents and Republicans reported less criticism of candidates in RCV cities. Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 13

Figure 8: Remember Candidates Criticizing Each Other, by Select Demographics, RCV and Plurality Cities 3.3 Perceived Negativity Overall Patterns In addition to improved satisfaction with local elections, residents of RCV cities were more likely to perceive the 2014 campaign as less negative than previous campaigns. Respondents were asked whether they believed the campaigns this year were more negative, less negative, or about the same compared to other recent local political contests. A follow-up question was asked if the respondent reported the campaigns were more or less negative about whether the campaigns were a lot or a little more (or less) negative. Combined these two questions created a 5-piont scale. The perception that the 2014 campaign was less negative than earlier local elections was more prevalent in RCV cities than in plurality cities. In RCV cities, 17% of all respondents thought that the 2014 campaign was less negative, five percentage points higher than in plurality cities (12%). Overall, 14% of respondents reported perceiving less negativity in the 2014 campaign compared to earlier contests. Across both city types, young respondents and those with a high school diploma or less were most likely to perceive the 2014 contest as less negative than in other recent local contests (Figure 9). Twenty-one percent of 18 29 year old respondents and 17% of respondents who did not attend any college reported less negativity in 2014. Interestingly, the perception of less negativity was constant across income groups. Democrat respondents were more likely to believe the 2014 campaign was In RCV cities, 17% of respondents thought the local 2014 campaign was less negative than in previous years. less negative than were their Independent counterparts (Figure 10). Among race and ethnic groups, Asian respondents were the least likely to report that the 2014 campaign was less negative than in the past, although this was not statistically significant. This forms the Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 14

baseline for comparison when viewing Figures 11 and 12, which break out these general patterns by respondents living in RCV and plurality cities. Figure 9: Perceived Less Negativity than Recent Local Election, by Age, Education and Income, All Surveyed Cities Figure 10: Perceived Less Negativity than Recent Local Election, by Select Demographics, All Surveyed Cities Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 15

Comparing RCV and non-rcv Cities When we compare RCV and plurality cities (Figures 11 and 12), we find that most socioeconomic and demographic groups reported less negative campaigning in RCV cities than their counterparts in plurality cities. Indeed, only those respondents with a high school education or less did not report lower levels criticism in RCV cities than in plurality cities. Across all age groups and income groups citizens reported less negativity in RCV cities than plurality cities. The consistent pattern across these demographic subgroups suggests campaign negativity was lower in the RCV cities than in comparable cities with plurality elections. However, in RCV cities, respondents who attended college were more likely to perceive less campaign negativity than respondents who did not. The opposite is true in plurality cities. Only Republicans and respondents who did not attend college reported a greater reduction in negativity in plurality cities than in RCV cities. Higher proportions of respondents from all ethnic and racial groups reported that the 2014 election campaigns were less negative in RCV cities than in plurality cities (Figure 12). Importantly, African American respondents in RCV cities were significantly say the campaigns were less negative than black respondents in plurality cities. Independents were also significantly more likely to say the campaigns were less negative in RCV cities than in non-rcv cities. Figure 11: Perceived Less Negativity than Recent Local Election, by Age, Education and Income, RCV and Plurality Cities Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 16

Figure 12: Perceived Less Negativity than Recent Local Election, Select Demographics, RCV and Plurality Cities Overall, likely voters responses to the three questions about the tone and conduct of candidate campaigns for local offices in their city in 2014 are revealing. They show that, under RCV cities, likely voters: (1) are more satisfied with the conduct of candidates; (2) witness less candidate criticism of other candidates; and (3) notice a reduction in negative campaigning. It is particularly striking that these impressions cut across socioeconomic and demographic groups, such that virtually every sub-group s experience is better under RCV. Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 17

4. Voter Understanding of Ballot Instructions, Voting Systems and the Top-Two Primary Respondents were asked whether they voted in the most recent (November 2014) local election. Those who reported voting were queried about how easy it was to understand the ballot instructions. Additionally, respondents in RCV cities were asked about how well they understood ranked choice voting (RCV), while respondents in plurality cities were asked about their understanding of plurality using parallel question wording. All respondents were also questioned about their comprehension of the Top-Two Primary used in primary elections in California. Voters reported high and equivalent levels of understanding of the two voting systems used for their respective city elections at the broad level. Respondents reported slightly higher levels of understanding the ballot instructions in plurality cities than in RCV cities; which likely reflects that RCV is somewhat of a newer process. But, overall, 9 in 10 respondents reported it was somewhat or very easy to understand the instructions on their ballots in both RCV and non RCV cities. Overall, African Americans and lower income respondents had In the 2013 Eagleton Poll, over 90% of respondents in RCV cities reported that understanding the instructions on their RCV ballot was either somewhat or very easy. more difficulties understanding their voting instructions. The most robust finding about the comprehension of ballot instructions is that the disadvantage that African Americans faced in understanding ballot instructions in plurality cities was reversed in RCV cities, suggesting that well-worded instructions matter more to understanding than does the detail of the voting system. Voters in RCV cities were more likely to have a detailed knowledge of RCV than voters in plurality cities were to have a detailed understanding of plurality. In both RCV and plurality cities, levels of understanding of voting systems were positively correlated with increasing levels of education. Of all groups surveyed, Republicans in RCV cities reported the highest level of understanding of their city s voting system. More respondents reported understanding RCV extremely well or very well than reported the same level of understanding of the Top-Two Primary. 4.1 Ease of Understanding Voting Instructions Overall Patterns Those respondents who reported voting in their local election in 2014 were asked, When you voted in the recent election, how easy was it to understand the voting instructions? Across all surveyed cities, the vast majority (91%) of voting respondents reported that it was very or somewhat easy to understand voting instructions. More voters (93%) in plurality cities reported that voting instructions were easy to understand than in RCV cities (88%). Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 18

Figure 13: Ease of Understanding Voting Instructions, by Age, Education and Income, All Surveyed Cities Figure 14: Ease of Understanding Voting Instructions, by Select Demographic, All Surveyed Cities Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 19

Overall, understanding of ballot instructions was the most skewed against low income voters (Figure 13) and African-American voters (with just 80% reporting ballot instructions were easy to understand - Figure 14). Lower levels comprehension of ballot instructions was also reported for less educated voters and the youngest (under 30 years) and eldest (over 65 years) voters. However, across demographic and socioeconomic groups over 80% of respondents indicated they understood the ballot instructions with ease. Comparing RCV and non-rcv Cities In RCV cities, respondents from all socio-economic groups found the voting instructions easy to understand (Figure 15) with over 80% of every group bar one reporting the ballot instructions were easy to understand. The youngest voters (those under 30 years) reported the lowest levels of understanding the ballot instructions (79%) (Figure 15) compared to 94% of those age 30 49 year old. This suggests that clear ballot instructions really matter to inexperienced voters. In plurality cities, understanding the ballot instructions was heavily skewed in favor of high income respondents. In RCV cities, lower income voters (especially those whose family income was less than $20,000) reported relatively high levels of understanding ballot instructions. Interestingly too, respondents in RCV cities who had never attended college reported it was easy to vote: significantly higher than those in RCV cities who had attended college and higher than those in plurality cities who had never attended college (Figure 15). These findings indicate the power of voting instructions and voter education to overcome traditional disparities in electoral participation. Figure 15: Ease of Understanding Voting Instructions, by Age, Education and Income, RCV and Plurality Cities Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 20

In RCV cities, 90% of African-American respondents reported that understanding their ballot instructions was easy, compared to only 65% in plurality cities. FairVote: The Center for Voting and Democracy (April 2015) Reported understanding of ballot instructions was less stratified by race, gender and marital status in RCV cities than in plurality cities. In particular, in RCV cities, the disparities between different racial groups and comprehension of ballot instructions were less than in plurality cities (Figure 16). A significantly higher percentage of African-American respondents reported that ballot instructions were easy to understand in RCV cities (90%) than in plurality cities (65%). The difference between the two percentages adds evidence of the importance of the phrasing of ballot instructions. In particular, the difference may indicate that careful and clear ballot design may be far more important to understanding than the mechanics of the voting system used in a jurisdiction. Asian Americans again stand out, once again, in the RCV cities for their relatively low levels of understanding the ballot instructions (82%), reflecting their lower rates of participation in politics. However these patterns were not significantly lower than other racial and ethnic groups. Figure 16: Ease of Understanding Voting Instructions, by Select Demographics, RCV and Plurality Cities 4.2 Understanding RCV and Plurality Respondents in all cities were asked about their understanding of the electoral system used in their city. Correspondingly, respondents in RCV cities were asked, Overall, how well do you think you understand ranked-choice voting?, while respondents in plurality cities were Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 21

asked, Overall, how well do you think you understand plurality voting? Levels of understanding of the electoral system used in RCV and plurality cities were comparable. In RCV cities, 84% reported they understood RCV extremely, very or somewhat well. In plurality cities, 83% of respondents reported they understood plurality extremely, very or somewhat well. If we consider the comprehensiveness of respondents reported understanding of voting systems, we find that understanding of RCV is actually higher than understanding of winner take all elections in plurality cities. In both RCV cities and plurality cities, 49% of voters reported understanding their local voting systems extremely well or very well. However, in RCV cities, 22% of respondents reported they understood RCV extremely well, while only 12% of respondents in plurality cities reported understanding plurality extremely well. Similar socioeconomic and demographic patterns existed in both RCV and plurality cites relative to voter comprehension of electoral systems. Under both systems, respondents understanding of the voting system used in their local elections was correlated with their level of education, with less educated respondents reporting lower levels of understanding of both RCV and plurality systems (Figures 17 and 18). In plurality cities, increasing age was weakly associated with increasing understanding of plurality voting (Figure 18). By contrast, in RCV cities, voters aged between 30 and 49 reported the most understanding of RCV (Figure 17). In neither case were the differences statistically significant. In both RCV and plurality cities, respondents with a family income between $75,000 and $150,000 reported the highest levels of understanding of their local voting systems. Figure 17: Understanding of Ranked Choice Voting, by Age, Education and Income, RCV Cities. Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 22

Figure 18: Understanding of Plurality Voting, by Age, Education and Income, Plurality Cities. In both RCV and plurality cities, over 85% of African-American and Latino respondents reported understanding their city s respective voting systems (Figures 19 and 20). Only 75% of Asian Americans and Independent respondents reported understanding RCV, suggesting the need for targeted voter education. Almost 90% of Republicans in RCV cities reported understanding RCV, the highest level reported by any group. By contrast, in plurality cities, only around 78% of Republican respondents reported understanding plurality voting (Figure 20). This difference is statistically significant at the 95% confidence level. In RCV cities, more than 85% of African- American and Latino respondents and 90% of Republican respondents understood RCV. Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 23

Figure 19: Understanding of Ranked Choice Voting, by Select Demographics, RCV Cities. Figure 20: Understanding of Plurality Voting, by Select Demographics, Plurality Cities. Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 24

4.3 Understanding RCV and Top-Two FairVote: The Center for Voting and Democracy (April 2015) In addition to their understanding of their city s voting ballots and electoral system, respondents were asked about their understanding of the Top-Two Primary system. The Top-Two Primary was adopted by voters in California in 2010 for all state and congressional offices and used in 2012 and 2014. Levels of understanding of RCV and the Top-Two Primary were relatively even at the broadest level, but RCV was better understood than the Top-Two Primary system. In plurality cities, 87% of respondents reported understanding the Top-Two Primary extremely, very or somewhat well. In RCV cities, 85% of respondents reported understanding the Top-Two Primary at least somewhat well, while 84% reported the same level of understanding of RCV (Figure 21). These differences are not statistically significant. However, respondents in RCV cities had a more comprehensive knowledge of RCV. While only 41% of respondents in plurality cities and 40% of respondents in RCV cities understood Top-Two at least very well in RCV cities, significantly more (49%) respondents in RCV cities understood RCV very well or better. Theese differences are statistically significant (Figure 22). Simlarly, in RCV In RCV cities, 40% of respondents understood Top-Two extremely or very well, while 49% understood RCV extremely or very well. cities, 22% of respondents reported understanding RCV extremely well, while 18% reported understanding Top-Two extremely well. In Oakland, respondents had a much better grasp of RCV than Top-Two. Eighty-four percent reported understanding the Top-Two Primary at least somewhat well, while 88% understood RCV to that same degree (Figure 21). When we consider those who understood the institutions more comprehensively, the depth of understanding of RCV is especially apparrent. Only 38% of Oakland residents, for example, understood Top-Two extremely or very well; significantly more (51%) understood RCV extremely or very well (Figure 22). Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 25

Figure 21: Understanding the Top-Two Primary and Ranked Choice Voting Extremely, Very or Somewhat Well, All Surveyed Cities, Plurality Cities, RCV Cities and Oakland 90% 85% 80% 75% 70% 65% 60% All cities: Top-Two Non-RCV Cities: Top- Two RCV cities: Top-Two Oakland: Top-Two RCV Cites: RCV Oakland: RCV Figure 22 Understanding the Top-Two Primary and Ranked Choice Voting Extremely or Very Well, All Surveyed Cities, Plurality Cities, RCV Cities and Oakland 60% 50% 40% 30% 20% 10% 0% All cities: Top- Two Non-RCV Cities: Top Two RCV cities: Top-Two Oakland: Top-Two RCV Cites: RCV Oakland: RCV While respondents reported slightly higher levels of ease understanding ballot instructions in plurality cities than in RCV cities, it is clear that RCV is understood as well and perhaps better than other voting systems and electoral innovations like plurality and the Top-Two Primary. Greater understanding is especially clear when we focus on more comprehensive understandings of these electoral institutions. Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 26

5. Support for RCV in Californian Cities In the 2013 survey, over 60% of respondents in RCV cities supported the use of RCV in local elections. In the 2014 Rutgers-Eagleton poll, respondents were asked whether they thought ranked choice voting, where voters can rank candidates in order of preference with their first choice counting most, should be used in local elections in their city. The survey revealed support for RCV across most demographic groups in both RCV and plurality cities. Minorities, younger voters, Independents (compared to Democrats and Republicans), and the less educated, in particular, prefer voting using preference voting or RCV. Among those with an opinion, ranked choice voting was supported by majorities in both RCV cities (57%) and plurality cities (54%). Greater proportions of younger, less educated, less affluent and middle class, Asian, and Latino respondents supported RCV than did older, higher income or more educated respondents (Figures 23 and 24). Overall, support for RCV was higher among Asian-American (70%) and Latino (59%) respondents than it was among African Americans (47%) and white non-hispanics (48%) (Figure 24). High support for RCV among Asian Americans and Latinos is important in that these groups have typically participated in politics at lower levels than other groups. There was no statistically significant difference in support for RCV by marital status or gender. Figure 23: Support for RCV by Age, Education and Income, All Surveyed Cities* * As a percentage of those with an opinion (excludes don t knows and refused). See Methodology Section for more information. Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 27

Figure 24: Support for RCV by Select Demographics, All Surveyed Cities * * As a percentage of those with an opinion (excludes don t knows and refused). See Methodology Section for more information. Figure 25: Support for RCV by Age, Education and Income, RCV and Plurality Cities* * As a percentage of those with an opinion (excludes don t knows and refused). See Methodology Section for more information. Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 28

Figure 26: Support for RCV by Select Demographics, RCV and Plurality Cities* * As a percentage of those with an opinion (excludes don t knows and refused). See Methodology Section for more information. Experience appears to cement support of RCV by groups inclined to RCV in the abstract: the tendency for younger, Asian, Latino, lower income or less educated respondents to support RCV was stronger among those in RCV cities as it was in plurality cities (Figures 25 and 26). Interestingly, around 72% of Asian-Americans respondents in RCV cities supported the use of RCV, even though they reported lower levels of satisfaction with the conduct of candidates campaign (Figure 4) and comprehension of RCV (Figure 20). Strong support for RCV among this group is notable, and indicates they may see RCV as an alternative mechanism to seek representation in local elections. Overall, support for RCV is broad and deep. A considerable majority of respondents in RCV cities are content with their status quo RCV and that support is strongest among groups traditionally poorly served by democratic institutions: the young, low-income and minority voters. Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 29

6. Conclusion The survey paints a complex but positive picture of voter experiences with, and support of, RCV. As we have seen, likely voters in cities that used ranked choice voting (RCV) in their local elections were indeed more satisfied with the conduct of candidate campaigns, and perceived significantly less candidate criticism and negative campaigning in the lead up to the November 2014 election. Virtually all demographic groups studied reported less negativity and less candidate criticism in elections conducted using RCV than in elections conducted using plurality voting methods. On the basis of this expansive and independent survey, and the control group of plurality voting cities, it appears RCV is conducive to a less negative and adversarial campaign season for all socioeconomic and demographic groups, which, naturally, corresponds to a more satisfied electorate. While voter understanding of the instructions on RCV ballots was ever so slightly lower than in plurality ballots, it was very high in all cities with over 80% of respondents reporting they understood the voting instructions across demographic groups in both RCV and plurality cities. The demographic findings indicate that voter understanding depends more on the careful wording of ballot instructions than it does on the voting system used. This conclusion is especially apparent given the wide disjunction between African-American voter ease of understanding ballot instructions in plurality cities and in RCV cities. Likely voters understood RCV in greater detail than they understood either plurality voting or the Top-Two Primary. Likely voters often reported understanding plurality and Top-Two at the vague level ( somewhat well ). However, respondents less often reported understanding plurality and the Top-Two Primary comprehensively ( very or extremely well). Contrary to the claims of some commentators, likely voters appear to have less difficulty fully comprehending RCV than winner-take-all rules or the much anticipated Top- Two Primary. A majority of citizens support use of RCV in local elections. In both RCV and plurality cities, more voters supported the use of RCV than opposed it. Importantly, the greatest support of RCV came from traditionally under-represented communities including people of color, young people, and low-income voters. Especially striking is that the results of this study are consistent with the 2013 Rutgers- Eagleton poll that included identical survey questions. 5 That survey was administered in a different set of cities, all of which had local elections in November 2013, in the Midwest and East Coast. The 2013 findings, that in RCV cities perceptions of negativity and criticism were lower, satisfaction with campaigns was higher, ballot instructions and RCV were as easy to understand as plurality and RCV had majority support across most demographic groups, are replicated in this 2014 study of racially and ethnically diverse Californian cities. In two very difference sets of cities, the same effects were associated with the use of RCV in both 2013 and 2014. 5 See: Andrew Douglas (April 2014) Ranked Choice Voting and Civility: New Evidence from American Cities Ranked Choice Voting Civility Project Research Report No.1. Available at: http://www.fairvote.org/assets/rcv- Civility-Project/Ranked-Choice-Voting-Civility-Study-April-2014.pdf Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 30

Appendix: Methodology FairVote: The Center for Voting and Democracy (April 2015) For each of the seven US cities that used RCV elections in 2013 and 2014, multiple comparative cases cities that were demographically similar to RCV cities and were scheduled to hold elections (using plurality) on the same date were identified and used to create the survey sample. In 2013, three cities held RCV elections (Minneapolis and St. Paul, Minnesota and Cambridge, Massachusetts) on 5 November 2013. The seven plurality control cities samples were in the Midwest, Washington state and Massachusetts. All four cities holding RCV elections on 4 November 2014, and the seven control cities holding plurality elections were in California. RCV cities and their respective control cities are listed below in Table A.1. Table A.1: RCV Cities and Control Cities for Telephone Survey Sample RCV cities Matched plurality cities 2014 Survey (elections 4 November 2014) Berkeley (n=114) Alameda* (n=101) San Francisco (n=151) San Jose* (n=203) San Leandro* (n=395) Richmond* (n=349) Oakland * (n=685) Anaheim* (n=100) Santa Clara* (n=147) Santa Ana* (n=100) Stockton (n=111) 2013 Survey (elections 5 November 2013) Cambridge, MA (n=202) Lowell, MA (n=100) Worcester, MA (n=100) Minneapolis, MN* (n=812) Boston, MA* (n=268) St. Paul, MN (n=203) Seattle, WA* (n=270) Tulsa, OK* (n=269) Cedar Rapids, IA (n=108) Des Monies, IA (n=100) Notes: *City had a competitive mayoral election Number of respondents per jurisdiction in parentheses. Post-election, the Rutgers-Eagleton Institute of Politics conducted random registered voter listed sample-based telephone surveys (to both landlines and cellphones, in English and in Spanish) in the selected RCV and plurality cities. In November 2013, 2432 likely voters were surveyed: 1,217 in RCV cities and 1,215 in plurality cities. In November 2014, the telephone survey was fielded November 6-18, 2014 and screened for only self-reported registered voters. Interviews were terminated if respondents indicated they rarely follow local news or were otherwise uninterested in local elections. Target sample size was 2,400 respondents. Actual respondent count totaled 2,456 registered voters in eleven cities. In the four cities that utilize ranked choice voting rules in local elections for mayor and city council members Berkeley, Oakland, San Leandro, and San Francisco Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 31

1,345 respondents were surveyed. In the seven control cities that utilize plurality voting rules in local elections for mayor and city council members Alameda, Anaheim, Richmond, Stockton, San Jose, Santa Ana, and Santa Clara 1,111 respondents were surveyed (Table A.1). Quota sampling was set for each city in order to gather a roughly equal number of respondents from both RCV and non-rcv cities. 6 Larger samples were typically taken in cities holding competitive city-wide mayoral elections than in cities holding council elections in a handful of wards (Table A1). Survey weights were used for the analysis that account for differences in demographic distributions among survey respondents with demographic distributions among the population they are drawn from. Separate survey weights were used for analyses of RCV cities, non-rcv cities and the combined sample of RCV and non-rcv cities. In addition, cityby-city breakups were weighted using a city-specific weight. Each weight is constructed using race, age and education variables. For more information on survey methodology, please contact Professor Caroline Tolbert at the University of Iowa. 6 Quotas for each city are as follows: Alameda (100), Anaheim (100), Berkeley (100), Oakland (650), Richmond (350), Stockton (100), San Leandro (400), San Francisco (150), San Jose (200), Santa Ana (100), Santa Clara (150). Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 32

About the Ranked Choice Voting Civility Project The Rutgers-Eagleton Institute of Politics survey data presented here are part of a larger project investigating ranked choice voting in local elections. The project is funded by the Democracy Fund, conducted with Professors Caroline Tolbert, Todd Donovan, Martha Kropf, and David Kimball, and assisted by FairVote s staff and its Ranked Choice Voting Academic Advisory Committee. Professors Tolbert and Donovan are responsible for developing and evaluating the telephone survey. Tolbert is a professor of political science at the University of Iowa, where she focuses on voting, elections, and public opinion. She is the co-author of eight books and nearly 50 articles in academic journals. Donovan is a professor of political science at Western Washington University, where he studies the relationship between representation and electoral systems, public opinions, political behavior, direct democracy, and American state politics. He is the author of 10 books on elections and campaigns, and over 50 peer reviewed journal articles. Sarah John is a 2014-2015 Research Fellow at FairVote. She has a Ph.D. and has authored several academic articles on electoral systems and history. For more information and media enquiries please contact: Caroline Tolbert, Ph.D. Sarah John, Ph.D. Department of Political Science FairVote Democracy Fellow University of Iowa Takoma Park, Maryland Phone: 319-335-2358 Phone: 301-270-4616 Email: caroline-tolbert@uiowa.edu Email: sjohn@fairvote.org Our thanks to the Democracy Fund for financial support for this research project. Socioeconomic and Demographic Perspectives on Ranked Choice Voting in the Bay Area 33